A communication-aware framework for robotic field estimation

In this paper, we consider the problem where a fixed fusion center utilizes a number of mobile sensors in order to estimate the spatial variations of a field. The sensors measure the variations of the field in regions around their current positions and send their sensory data back to the fixed fusion center, by communicating over realistic fading wireless channels. The goal is to maximize the estimation performance at the fusion center, while maintaining the connectivity of the mobile sensors to it. In order to achieve this, we propose a localized gradient-based exploration strategy, which is based on switching between three modes of operation. The proposed approach is aimed at maintaining the connectivity of the mobile sensors and exploring the entire connected region, in the presence of realistic channels that experience path loss and fading. Our simulation results confirm the effectiveness of our proposed framework.